On the theory of the ellipse of elasticity as a natural

Transcription

On the theory of the ellipse of elasticity as a natural
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
On the theory of the ellipse of elasticity as a natural discretisation
method in the design of Paderno d’Adda Bridge (Italy)
R. Ferrari & E. Rizzi
Dip.to di Progettazione e Tecnologie, Facoltà di Ingegneria, Università di Bergamo, Dalmine (BG), Italy
ABSTRACT: The Paderno d’Adda Bridge, Lombardia, northern Italy, is one of the very first great iron constructions designed through the practical application of the theory of the ellipse of elasticity, a graphical-analytical
method of structural analysis that was developed in the 19th century. It embeds a natural discretisation of the
structure into a series of elastic elements, treated then with standard tools of geometry of masses. In this work,
the application of such theory to the calculation of the parabolic arch of the bridge is inquired, attempting to
breathe, at the same time, the beauty of the architectonic and structural conception directly linked to that; later,
results are compared with much modern approaches that also consider now-available numerical discretisation
methods. A further, definite aim of this work is also that of trying to promote interest on the bridge, on its actual
state of conservation and future destinations. Not only it represents a true industrial monument and a living testimony of the scientific and technological developments of the time but also a beautiful, effective achievement
of architecture and engineering through the methods of Strength of Materials.
1
INTRODUCTION
1.1 The Paderno d’Adda Bridge
The Paderno d’Adda Bridge, also called San Miche-le
Bridge, is a metallic viaduct that crosses the Adda river
between Paderno and Calusco d’Adda to a height of
approximately 85 m from water, allowing to connect
the two provinces of Lecco and Bergamo, near Milano,
in the Lombardia region, northern Italy (SNOS 1889,
Nascè et al. 1984). At that location the river flowsdown from the exit of Lecco’s branch of Como’s lake
to the river Po through an impressive natural scenery
that even seems to have inspired celebrated paintings
by Leonardo (Fig. 1).
The main upper continuous beam, 5 m wide, is
formed by a 266 m long metallic truss supported by
Figure 1. Front view of the Paderno d’Adda Bridge,
1887–1889 (from up-stream; left bank Calusco, right bank
Paderno). A crossing train is visible inside the upper horizontal continuous beam. Automotive and pedestrian traffic
runs on top.
nine bearings. Four of these supports are provided
by a marvellous doubly built-in parabolic metallic
arch of about 150 m of span and 37.5 m of rise. The
bridge shares its architectural style with similar arch
bridges built in Europe at the time (Timoshenko 1953,
Benvenuto 1981, Nascè et al. 1984), like e.g. that of
Garabit (1884, France, Eiffel and Boyer) and Maria Pia
(1887, Oporto, Eiffel and Seyrig), both doubly hinged
at the shoulders, and the Dom Luiz I (1886, Oporto,
Seyrig), doubly built-in as that of Paderno. The viaduct
was quickly constructed between 1887 and 1889 (thus
practically at the same time of the most celebrated
Tour Eiffel), to comply with the needs of the rapidly
growing industrial activities in Lombardia. It was built
by the Società Nazionale delle Officine di Savigliano
(SNOS), Cuneo, Italy, under the technical direction
of Swiss Engineer Giulio Röthlisberger (1851–1911),
the man whom the design of the bridge is normally
attributed to. He was formed at the Polytechnic of
Zürich, graduated in 1872 and got later in charge of the
Technical Office of the SNOS since 1885, for 25 years.
The bridge is still in service, with alternated oneway automotive traffic, restricted to no heavy-weight
vehicles, and trains crossing at slow speed.
The bridge is designed through the graphicalanalytical methods of structural analysis that were
booming in the 19th century (Culmann 1880,
Timoshenko 1953, Benvenuto 1981). Specifically, it is
a remarkable application of the so-called theory of the
ellipse of elasticity (Culmann 1880, Belluzzi 1942).
583
This theory was originally conceived by Karl Culmann
(1821–1881) and then systematically developed and
applied by his pupil Wilhelm Ritter (1847–1906). It
represents a very elegant method for the analysis of
the flexural elastic response of a structure and is based
on an intrinsic discretisation of a continuous beam in
a series of elements, each with a proper elastic weight,
directly proportional to its length and inversely proportional to its bending stiffness. The theory of the
ellipse of elasticity is based on the concepts of projective geometry, which lead to a correspondence between
the ellipse of elasticity of the structure and the central
ellipse of inertia of the distribution of the elastic weight
of the structure. This correspondence brings back the
problem of the determination of the flexural elastic
deformation of a beam to a problem of pure geometry of masses, of more convenient solution and direct
interpretation in terms of the design of the structure.
1.2
Main technical features of the bridge
The main technical features of the bridge are reported
in details in Nascè et al. (1984), which is, to our knowledge, the most comprehensive publication, and one of
the very few, concerning the bridge. We rely very much
on this very valuable contribution and on the Technical Report (SNOS 1889) that was originally issued
at the time of the first try-out. Here, the essential
characteristics are reported.
The 266 m long upper flyover is made by a continuous box girder with nine equally-distributed supports,
at 33.25 m distance from each other. Four of the supports are sustained by a big parabolic metallic arch; two
of them bear directly on the same arch’s masonry abutments (made with Moltrasio masonry, with Baveno
granite coverings); a seventh, on the Calusco bank,
rests on a smaller masonry foundation placed between
the arch shoulder and the higher bridge supports; the
last two, in masonry work as well, are the two direct
beam bearings at its two ends, on top of the two river
banks. The four piers resting on the arch are placed
symmetrically, in between keystone, haunches and
shoulders of the arch.The inner side of the beam girder,
on which the railway is located, runs at about 255.00 m
on the sea level (osl); the rails are placed at 255.45 m
osl, the upper road at 261.75 m osl. The main vertical longitudinal trussed beams of the upper continuous
girder are 6.25 m high and placed at a respective transverse distance of 5.00 m, leaving a free passage for
the trains of 4.60 m. They are composed of two main
T-ribs connected by a metallic truss. The upper-level
road is 5.00 m wide and includes also two additional
cantilever sidewalks, each 1 m long, with iron parapets
1.50 m high.
The big arch is composed by two couples of secondary inclined arches. Each couple is formed by two
arches posed at a respective distance of 1 m and laying symmetrically to a mean plane inclined of about
±8.63◦ to the vertical. The parabolic axis of the arch
has a span of 150.00 m and rise of 37.50 m in the
inclined plane; the transverse arch’s cross section is
4.00 m high at the keystone and 8.00 m high at the
abutments (i.e. in the same 1:2 ratio between rise and
half span). The two mean inclined planes of the arches
are located at a distance of 5.096 m at the keystone and
16.346 m at the shoulders. The wall of each composing
arch is also a truss structure with two main T-ribs connected by vertical and inclined bars. The two couples
of twin arches are gathered together by two transverse
brace systems located at the extrados and at the intrados of the arch’s body. In essence, the resulting cross
section of the main parabolic arch supporting the horizontal beam is trapezoidal, with variable, increasing
cross section from the crown to the shoulders. This,
and specifically the inclination of the twin arches, is a
beautiful key feature of Röthlisberger’s conception of
the bridge, in view of counteracting effectively wind
and transverse horizontal actions in spite of the considerable slenderness of the structure. The arch cross
section at the impost is inclined of 45◦ to the horizontal, so as the local tangent to the parabolic axis
of the arch to the vertical. The vertical bridge piers
that sustain the upper continuous beam are made by
eight T-section columns, linked to each other by a brace
system with horizontal bars and St. Andrew’s crosses
and, on top, by transverse beams that directly serve as
supports for the bearing devices of the upper beam.
For inspection and maintenance purposes a 1 m large
boardwalk is provided into the body of the arch and,
inside the bridge piers, a system of ladders along their
height. The bridge is a riveted wrought iron structure of
about 2600 t of metals, with near 100000 rivets just in
the arch.
1.3 Aim of this work
In this work, which in its main part largely refers to
a study developed in a Laurea Thesis (Ferrari 2006),
a detailed analysis of the SNOS Report (1889) is presented. The point of view here is the following: inquire
the application of the theory of the ellipse of elasticity to the calculation of the bridge, breathe the beauty
of the architectonic and structural conception directly
linked to that, compare results with modern structural
approaches that also consider now-available numerical
discretisation methods.
After a careful review of the Report by the SNOS,
a full 3D truss Finite Element model of the arch of
the bridge has been elaborated, based also on direct
inspections of the bridge and on the screening of
the marvellous original drawings that are guarded
at the Archivio Storico Nazionale di Torino. Different loading conditions have been considered and
results compared with those reported in the SNOS
Report, showing the remarkable accuracy of the
adopted graphical-analytical methods and allowing to
584
experience the unrepeated beauty of the original analysis with respect to rather impersonal computer structural analysis. Moreover, the model that has been put
in place shows promise for possible further analyses
that could inspect other behaviours of the bridge, as for
example dynamical and inelastic, also connected to the
present and future state of conservation of the structure. These aspects are left for further developments
of the present study.
The paper is organized as follows: Section 2 provides a short account on the theory of the ellipse
of elasticity; Section 3 reports its application to the
structural analysis of the arch of the bridge; Section 4 presents an independent validation of the original design results with present analytical-numerical
methods.
2
2.1
ON THE THEORY OF THE ELLIPSE OF
ELASTICITY
Fundamentals
The theory of the ellipse of elasticity can be considered as a main icon of the so-called Graphical Statics,
the discipline which often characterised the resolving
approach of practical design problems during the 2nd
half of the 19th century. It represents a very elegant
and practical method for the analysis of the flexural
response of an elastic structure. It is based on an intrinsic discretisation of a continuous elastic problem. This
theory is basically associated to the two outstanding
figures of Culmann and Ritter, but also of people, like
Giulio Röthlisberger, that were formed at the time at
the Polytechnical Schools in Europe and that became
later structural engineers and designers and largely
contributed in the practical and effective application
of the method.
The theory is based on the following main hypotheses (we refer here to the Italian text by Belluzzi 1942,
which reports results from the technical literature
of the time, basically ascribed to the two names of
Culmann and Ritter): (a) linear elastic behavior of
the material and the structure, which leads to the
proportionality between acting forces and (reversible)
displacements provoked by them (property that in turn
implies the validity of the principle of superposition
of effects); (b) existence of the ellipse of elasticity,
referred to a section of a structure; (c) correspondence
between the latter and the central ellipse of inertia of
the distribution of the so-called elastic weight of the
structure.This correspondence transforms the problem
of the determination of the elastic response of a continuous structure to a task of pure geometry of masses.
The latter can be feasibly handled by taking advantage
of the assumed discrete character of the distribution of
the elastic weight and is endowed with a visible interpretation of the elastic performance of the structure,
Figure 2. Representation of a force R applied along line of
action r to a beam section A and the centre of rotation C of
A. C is the symmetric of C with respect to the centre S of the
ellipse of elasticity of beam AB referred to section A.
in view of its conception and design. Furthermore, the
ellipse itself may actually play the role of an hidden,
underlying, graphical construct. Indeed, the properties of projective geometry that are attached to that
allow the elastic solution of the structure even without
the explicit drawing of the ellipse itself. The methods
are also said graphical-analytical because, in practice,
main technical steps that are framed on the graphical constructions may be carried-out analytically, by
working-out formulas that arise from the inspection
of the drawings (Belluzzi 1942).
The concept of the ellipse of elasticity referred to a
section of an elastic structure is achieved by inspecting the correspondence existing between the line of
action r of a force R applied to a section A of a general, curvilinear, elastic beam (with little curvature and
continuously-varying cross section) and the centre of
rotation C of the same section (Fig. 2).
In particular, refer to a planar beam acted upon
by forces laying in the same plane and cross sections
of the beam that, during the beam’s deformation, are
assumed to remain plane and perpendicular to the geometric axis, also deforming in its original plane. The
theory states that there exists an involutory relationship
between the line of action r and the centre of rotation
C of the section. Moreover, the ellipse of elasticity
is the fundamental real conic of the polarity existing
between the line of action r and the point C , which is
the symmetric of C with respect to the centre S of the
ellipse. In other words, the ellipse of elasticity can be
defined as the fundamental conic with respect to which
the lines of action r and the respective centres of rotation C correspond to each other through an antipolarity
relationship.
The determination of the central ellipse of inertia of
the distribution of the elastic weight of the structure,
which coincides with the ellipse of elasticity above,
is linked first to the definition of the general concept
of elastic weight and then to the quantification of its
distribution for the structure under consideration. The
concept of elastic weight goes as follows. If on a section A of a beam, a moment M acts in the plane which
585
contains the geometric axis, it causes a rotation φ of
A, around the centre S of the ellipse of elasticity. The
rotation is proportional to the applied moment M as:
where G represents the so-called elastic weight of the
beam. Thus, G can be defined as the angle of rotation φ that is caused by the application of a unitary
moment M = 1; it depends on the beam’s geometrical
and physical properties; it gives a global measure of
the beam’s aptitude to deform. In case of a straight
cantilever beam of length l loaded by a moment M
at its free end, composed by a linear elastic material
with Young’s modulus E and endowed with a constant
cross section with moment of inertia J with respect to
the axis perpendicular to the beam’s plane, it turns out
that, referring to the case of flexure of de Saint Venant,
the rotation of the free end is:
This relation clearly illustrates the physical meaning of
G as the global parameter that expresses the flexural
elastic deformability of the structure.
Now, the idea arises of thinking at the structure as
the assembly of a series of discrete elastic elements of
length s, each with a proper elastic weight
such that the total elastic weight of the structure is
represented by the discrete distribution of these elastic weights. It is possible to demonstrate that such
a sought distribution of G is univocally known only
for a statically-determined structure, whereas for a
statically-undetermined structure the distribution of
elastic weights is not univocally defined. This is not
surprising, due to the redundancy of equilibrium in an
hyperstatic system. Despite this, an hyperstatic structure can still be solved, via the Forces Method (with
hyperstatic quantities as unknown), through the superposition of effects on underlying isostatic structures
and imposition of the corresponding compatibility
conditions. As the underlying isostatic structure can
also be analysed with a univocally-defined distribution of elastic weights, such distribution can also be
used to solve the original hyperstatic structure. Thus,
indirectly, its ellipse of elasticity can in essence be
determined, so the corresponding ellipse of inertia of
the distribution of elastic weights.
This allows one to write the so-called “theorems
of the theory of the ellipse of elasticity” (Belluzzi
1942), as a function of the properties of the distribution of elastic weights. For example, the rotation φ
Figure 3. Representation of the quantities reported in Eq. (4)
for the calculation of the rotation and displacements of terminal section A caused by a force Q applied to the same
section.
and horizontal and vertical displacements dx , dy of a
terminal beam section A caused by a force Q applied
to the same section along line of action q (Fig. 3), can
be nominally written as follows:
where G =
G represents the total elastic weight of
the structure; Sq , Jxq , Jyq the static moment of G with
respect to q and the centrifugal moments of inertia of
G with respect to q and axis x, and q and axis y. These
parameters depend only on the distribution of elastic
weights and on the position of the applied load Q, and
can be expressed as a function of the quantities xS , yS ,
uS , uX , uY depicted in Fig. 3.
The point S in Fig. 3 represents the centre of gravity of the elastic weights of the structure; the points
X,Y represent the antipoles of the reference system
axes x, y with respect to the central ellipse of inertia
of the elastic weights. It is then apparent that, once
the position of points S, X, Y and total elastic weight
G are found, the elastic response of the structure is
determined. The coordinates xS , yS , xY , yX defining
the position of these points can be evaluated by standard calculations of geometry of masses, once given
the discrete distribution of elastic weights.
2.2 Application to a doubly built-in
parabolic arch
In the SNOS Report (1889), the remarkable application of the theory of the ellipse of elasticity to the
analysis of the arch of the bridge refers to the determination of the elastic response of a parabolic arch,
built-in at the two extremities, which is subjected to a
vertical load P (that can be put equal to 1) and acting
in an arbitrary position along the arch, at a horizontal
distance a from left extreme A (Fig. 4).
First, the position of the line of action of the reaction A needs to be determined. This can be solved by a
586
Figure 4. Calculation of line of action LO of left reaction
A; the segments FL and VO that the left reaction A locates on
the vertical lines from A and on load P, below and above the
horizontal line from centre S are determined.
Figure 6. Equilibrium requires that the left and right reactions A and B form with load P a closed polygon of forces.
Segments V = OQ and V = QP represent the vertical components of A and B, with V + V = P; segment H = TQ their
horizontal component.
Alternatively, one may also proceed as sketched in
Fig. 4, by calculating parameters µ, ν and segments
FL, VO, with same results. The last relations in (5)
also give the magnitude of reaction A, given P.
Once the position of the lines of action of left builtin reaction A, and, by equilibrium, of right reaction
B are known, the value of their vertical components
V , V and (common) horizontal component H can be
found as follows:
Figure 5. Calculation of line of action HK of left reaction A;
segments a, b, c to be drawn at points S,X,Y are determined
analytically. Points H,K are then located, so the direction of A.
graphical-analytical procedure. Figs. 4 and 5 represent
two ways to solve this problem (SNOS 1889, Ferrari
2006). Underlying to these constructions lay the compatibility conditions φ = 0, dx = 0, dy = 0 for the left
built-in constraint at terminal section A. These can be
worked-out from Eq. (4), leading to:
where A is now taking the role that force Q had in
(4) and the superposition of effects is considered with
load P, so that uS = a, uX = b, uY = c (a, b, c denote
segments used below in Fig. 5) are quantities similar to
those entering Eq. (4), but related to load P at position a
from A. They can be calculated analytically as follows,
given the distribution of elastic weights:
Once scalars uS = a, uX = b, uY = c are evaluated, the
construction in Fig. 5 determines the position of A.
where quantities a, a , f, f are represented in Fig. 6.
Following similar arguments, it is possible to derive
the in-plane arch’s deflections dy at any position x of
the arch, for the different locations a of the load P
acting on the arch (SNOS 1989). The equation that
gives the mean parabolic line at the arch is taken as:
where f and l represent now the rise and span of the
arch. The profile of the arch is symmetric with respect
to the vertical axis x = l/2 at half span. The formulas
for dy obtained by the SNOS with this graphicalanalytical procedure correspond indeed to those that
may be obtained by the application of theVirtual Works
Principle (VWP):
where H , V , M represent the “components” of the
reaction force A (built-in moment M is positive clockwise) and C = EJdx/ds = cost is a quantity related to
587
determined without the constant proportionality factor E = 17000000 t/m2 , Young’s modulus of the iron,
i.e. G = s/J . The coordinates xS , yS , xY , yX and
total elastic weight G = EG are finally found as:
Figure 7. Discretisation of the half arch in 14 elements
with x = cost, with indication of the geometrical quantities useful to determine the corresponding elastic weights
(quotes in m).
the bending stiffness and local inclination of the arch.
C is assumed by the designer constant along the arch,
since while J decreases from the shoulder to the keystone, the ratio dx/ds between horizontal projection dx
of infinitesimal element length ds of the arch and ds
itself increases (SNOS 1889).
3
3.1
The constant C in Eq. (9) is also evaluated in
55539000 t · m2 .
Anyway, tables are presented in which the left reaction and the deflections of the arch are determined for a
unitary load located at the various elastic elements. In
practice, these influence coefficients, which are later
used for the design of the truss members, are determined by a true application of the theory of the ellipse
of elasticity as applied to the arch.
3.2
The SNOS Report analyses independently, one by
one, the various loadings on the arch, for subsequent
superposition of effects:
– permanent weight of the arch;
– permanent weight of the upper girder beam, of the
bridge piers and vertical actions induced by the
wind acting on the girder beam;
– accidental vertical load on the upper girder beam;
– temperature effects and compression on the arch
due to the horizontal thrust H ;
– direct horizontal wind action on the arch.
STRUCTURAL ANALYSIS OF
THE BRIDGE
Explicit application of the theory of
the ellipse of elasticity to the analysis
of the arch of the bridge
Going to the explicit definition of the distribution of
elastic weights made by the SNOS (1989), all the calculations reported there were developed considering
an arch which is the projection on their mean inclined
plane of one of the two couples of inclined parabolic
arches that are placed symmetrically to the vertical
longitudinal median plane of the bridge. Such arch is
placed on a plane inclined of about α = 8.63◦ to the
vertical (such that sin α = 0.15) and consists of a truss
beam with parabolic axis of 150 m of span and 37.5 m
of rise in such plane, having extrados and intrados lines
both described by parabolic functions so as to determine a cross-high of the arch of 4 m at the keystone
and of 8 m at the abutment. On the basis of this model,
the elastic weights of the structure have been calculated according to a symmetric structural discretisation
with 28 elements of different s extensions as reported
in Fig. 7.
In the Report, the procedure adopted to calculate
the elastic weights G = s/EJ is not really apparent. An attempt of careful analysis is provided in
Ferrari (2006). Also, the elastic weights are actually
Evaluation of the loads
For each listed item, the SNOS reports the calculation
of the stresses in the various arch’s elements, as well
as the final value that arises by the superposition of
effects.
3.3 Dimensioning of the structural elements
The Report by the SNOS provides the calculation
of the stresses in the various bar elements of the
arch and at the stone abutments as compared to the
target admissible values that are summarised below.
This is done for the final geometries of the structural members. On the other end, the Report does not
provide specific information about the design procedure that has lead, through pre-dimensioning, to such
final structural dimensions.As the overall architectural
and structural conception of the entire viaduct, these
phases seem to be linked to the engineering practice
and experience of the designer with the standards of
metallic carpentry in use at the time.
The material employed in the structural members
of the bridge is a wrought iron, with very low carbon percentage of about 0.01% (Nascè et al. 1984).
The admissible stresses are taken differently for each
structural component of the bridge: for the main upper
and lower arch’s ribs 6.0 kg/mm2 ; for the vertical and
588
Table 1. Arch’s vertical deflections calculated at design
stage for four different loading tests (SNOS 1889, p. 71).
Negative values indicate downward displacements.
Figure 8. Scheme with four test loading configurations,
with indication of the four piers resting symmetrically on
the arch (view from down-stream; Paderno left side, Calusco
right side). Pier III is at half length of the upper continuous
beam.
diagonal bars linking the inferior and superior ribs of
the arch 6.0 kg/mm2 ; for the elements that compose
the transverse bridge’s brace system counteracting
wind 4.2 kg/mm2 . For each arch’s stone abutment,
the allowable compression stress is assumed in nearly
31 kg/cm2 . Stresses are all found safely below these
target values.
The executive drawings, including the various cross
sections of the structural elements, the details of the
riveted joints and part of the graphical/analytical calculations are reported in 147 marvellous drawing tables.
They are still in an excellent state of conservation and
show the meticulous design of the complex structure
of the bridge. Often, the calculus is integrated in the
drawing itself, which testimonies the beautiful, intimate link between conception, structural analysis and
executive design.
Deflections
Pier I
(m)
Pier II
(m)
Vertex
(m)
Pier III
(m)
Pier IV
(m)
Test I
Test II
Test III
Test IV
+0.0033
−0.0016
−0.0040
−0.0068
−0.0010
−0.0080
−0.0101
−0.0064
−0.0045
−0.0097
−0.0094
−0.0010
−0.0108
−0.0080
−0.0022
+0.0035
−0.0066
−0.0016
+0.0027
+0.0031
Table 2. Arch’s vertical deflections measured in situ at the
try-out for the corresponding loading tests (SNOS 1889,
p. 71).
Deflections
Pier I
(m)
Pier II
(m)
Vertex
(m)
Pier III
(m)
Pier IV
(m)
Test I
Test II
Test III
Test IV
+0.0038
0
−0.0016
−0.0063
+0.0001
−0.0079
−0.0102
−0.0058
−0.0044
−0.0080
−0.0073
−0.0007
−0.0106
−0.0102
−0.0014
+0.0042
−0.0056
−0.0012
+0.0025
+0.0026
with speed up to 45 km/h. Transverse oscillations at
the keystone were recorded in less than 3.142 mm and
vertical deflections in less than 10 mm. According to
Nascè et al. (1984) a 2nd, final, try-out took place in
June 1892, with different modalities and applied loads
but with similar results.
4 VALIDATIONS OF THE REPORTS’ RESULTS
BY ANALYTICAL-NUMERICAL METHODS
3.4 Viaduct’s tests
The first viaduct’s tests took place from 12th to 19th
May 1889. The measured deflections for loading conditions conforming to those considered at design stage
were compared to the corresponding theoretical values. To this purpose, the tests were carried out in two
moments: first, the different road loads were obtained
by deposition of gravel on the upper deck; second, with
a uniformly distributed gravel load of 3.9 t/m all over
the road, 6 locomotives with tender, each of 83 t of
weight, corresponding to a distributed load of 5.1 t/m,
were displaced on the railway track following four
different loading configurations (Fig. 8). Tables 1–2
below list the calculated and measured vertical deflections. The good overall agreement doubtlessly shows
the validity of the structural approach used by the
SNOS at design stage, in species the effectiveness of
the theory of the ellipse of elasticity as applied to the
explicit calculation of the arch of the bridge.
A final test was also run with 3 locomotives and
30 wagons, loaded by gravel, for a total weight of the
convoy of 600 t. The train was let running 3 times,
4.1 Analysis of the elastic arch by the VWP
In order to compare the results presented by the SNOS,
the hyperstatic scheme of the doubly built-in arch was
solved by the Virtual Works Principle. Such analysis allowed to notice some little inconsistencies in the
SNOS results. First, the reactions A and B due to P = 1
at distance a from left support A were evaluated and
compared to the Report’s results. The match was not
perfect, particularly for the bending moments at the
extremes of the arch. The differences did not seem to
be due to transcription errors, since the values reported
by the SNOS look coherent with formulas and tables
presented within the text. Second, further validation
concerning the arch’s deflections was attempted, since
they also did not seem to be calculated through the
reactions that are listed in the Report (Ferrari 2006).
However, the final values reported by the SNOS correspond, to a good degree of accuracy, to the present
results obtained by the VWP (Tables 3–4). The values
reported inTables 3–4 refer to a precise load configuration (1st distribution, see Section 4.2.2 andTable 5) and
589
Table 3.
4.2
Bending deflections listed in SNOS (1889, p. 59).
Load
Point
Pier I
(m)
Pier II
(m)
Pier III
(m)
Pier IV
(m)
Pier I
Pier II
Pier III
Pier IV
Total
−0.0089
−0.0011
−0.0004
+0.0001
−0.0103
−0.0029
−0.0042
−0.0003
+0.0001
−0.0073
+0.0071
+0.0026
+0.0005
−0.0000
+0.0102
+0.0042
+0.0030
+0.0001
−0.0001
+0.0071
Currently, an attempt is made of building a full FE
model of the bridge. So far, a truss mesh of the arch of
the bridge has been developed and appropriate loading
conditions has been considered for validation of the
previous results. The FE analysis has been run with
the commercial code ABAQUS® .
Table 4.
Bending deflections calculated by the VWP.
Load
Point
Pier I
(m)
Pier II
(m)
Pier III
(m)
Pier IV
(m)
Pier I
Pier II
Pier III
Pier IV
Total
−0.00893
−0.00126
−0.00038
+0.00006
−0.01051
−0.00298
−0.00425
−0.00033
+0.00010
−0.00746
+0.00703
+0.00255
+0.00055
−0.00004
+0.01009
+0.00422
+0.00297
+0.00016
−0.00013
+0.00722
Table 5. Total vertical deflections (1st distribution) reported
by SNOS (1889, p. 60) and present results by the FE model.
Point
Vertical loads SNOS deflections FE deflections
(t)
(m)
(m)
Pier I
+340.6
Pier II +144.0
Pier III −18.5
Pier IV
+4.9
−0.0120
−0.0113
+0.0062
+0.0054
Structural analysis of the arch by the FEM
−0.015903
−0.013847
+0.010322
+0.007222
regard the vertical deflections (due to pure bending) at
the four piers due to forces acting at their locations.
The reason of the discrepancies above may be due
to the fact that the SNOS might have used in the final
Report also data concerning a preliminary project.
Indeed, the bridge’s layout was slightly modified in
the executive project, due to new requirements on the
railway trace that were posed by the Strade Ferrate
Meridionali, after checks on the Adda’s banks (Nascè
et al. 1984). It is therefore possible that some specific
data were referring to a previous project, while important global quantities, such as the deflections caused
by external actions, were indeed corresponding to the
final one. As a matter of fact, even if this Report has
been probably conceived to present to a general audience the main steps of the calculations, including a
very valuable account on the theory of the ellipse of
elasticity and on its explicit application to the analysis
of the arch, it is doubtlessly very concise and it obviously presents concepts and practical considerations
that may not be directly apparent to the contemporary
reader.
4.2.1 Structural model
The FE model consists of a 3D truss frame, reproducing as much as possible the actual arch geometry. It
consists of two planar parabolic trusses referring to
the in-plane geometry of the arch (see Fig. 7), placed
in two inclined planes (of ±8.63◦ to the vertical). The
inclined planes are placed at a distance of 5.096 m from
each other at the axis of the arches at the keystone.
The truss nodes are linked to each other through a
reticular system that corresponds to the actual bracing of the arch. To each bar of the model, a cross
section with equivalent geometrical characteristics is
attributed (area, principal moments of inertia, torsional
stiffness). Some approximations were made as regards
to the section’s attribution to the superior and inferior
arch ribs, which are made with a variable number of
longitudinal plates. Also, at the node junctions, there
are additional reinforcing plates, for local stiffening.
In spite of this, the model has been simplified with
bars of constant average cross section. The model is
comprised of 752 beam elements and 266 nodes. Builtin constraints are imposed at the nodes of the arch
shoulders.
4.2.2 Obtained results
First trial loading cases considered a unitary load (1 t)
applied at the piers and at the keystone, in view of
verifying the order-of-magnitude agreement with the
deflections that were listed in the Report and calculated
here by the VWP. After these preliminary checks, the
deflections were evaluated for five given load distributions with vertical loads acting at the four piers (SNOS
1889, p. 58–62).
A sample of these outcomes is given in Figure 9
for the 1st load distribution already considered in
Tables 3–4, with main results summarized in Table 5.
Such distribution considers spans 2–3 charged by a
uniformly distributed load of 9 t/m, leads to a maximum pressure on Pier I and is somehow similar to
Test IV considered in the try-outs (Fig. 8). The maximum compressive axial force at the intrados of the left
shoulder is found in 377.0 t, in good agreement with
the value 751.6/2 = 375.8 t calculated by the SNOS
(1889, p. 62). Agreement is also found for the horizontal thrust H and vertical reactions V , V at the
shoulders, with H = 185.8 t, V = 417.2 t, V = 53.8 t.
The FE outcomes confirm, to a quite good degree of
accuracy, the SNOS results, showing the true potential
590
bridge. All this shows the ingenuous, effective and
beautiful approach to the design of the bridge. One
might think at this once contemplating the giant still
standing there silently, serving since almost 120 years
of duty, with an actual state of conservation that actually poses serious questions about its future survival.
We should take care of it.
REFERENCES
Figure 9. FE deformed configuration of the arch for the
1st load distribution reported in Table 5 (amplification
factor = 500).
of the theory of the ellipse of elasticity as applied to
the elastic analysis of the arch of the bridge.
5
CLOSING REMARKS
An attempt to scrutinise in details the SNOS
Report (1889) has been made in view of breathing the
original conception of the design of the bridge and its
specific calculation through the elegant method of the
theory of the ellipse of elasticity. The intrinsic discretisation in elastic elements is remarkable and makes a
natural connection with discretisations that can now
be provided by FE codes. The beauty of the executive
project can be totally appreciated only by the complementary screening of the 147 drawing tables of the
Belluzzi, O. 1942. Scienza delle Costruzioni. Bologna:
Zanichelli.
Benvenuto, E. 1981. La Scienza delle Costruzioni e il suo
Sviluppo Storico. Firenze: Sansoni.
Culmann, K. 1880. Traitè de Statique Graphique. Paris:
Dunod.
Ferrari, R. 2006. Sulla concezione strutturale ottocentesca
del ponte in ferro di Paderno d’Adda secondo la teoria
dell’ellisse d’elasticità. Laurea Thesis in Building Engineering. Advisor E. Rizzi, Università di Bergamo, 228
pages.
Nascè, V., Zorgno, A.M., Bertolini, C., Carbone, V. I.,
Pistone, G., Roccati, R. 1984. Il ponte di Paderno: storia e struttura – Conservazione dell’architettura in ferro.
Restauro, Anno XIII, n. 73–74, 215 pages.
Società Nazionale delle Officine di Savigliano 1889.Viadotto
di Paderno sull’Adda (Ferrovia Ponte S. Pietro-Seregno).
Torino: Tip. e Lit. Camilla e Bertolero.
Timoshenko, S.P. 1953. History of Strength of Materials. New
York: McGraw-Hill.
591